[Paper Review] Distinguishing Views in Symmetric Networks: A Tight Lower Bound

Dariusz Dereniowski, Adrian Kosowski+1 more|arXiv (Cornell University)|May 16, 2014

Graph theory and applications19 references5 citations

TL;DR

This paper establishes a tight lower bound of Ω(D log(n/D)) on the depth required to distinguish views of two nodes in symmetric, anonymous, port-labeled networks, proving that this depth is necessary to detect differences in their infinite views. The authors construct a family of networks where nodes have identical truncated views up to this depth but different full views, using a two-step construction involving base graphs with logarithmic depth symmetry and edge subdivision to scale diameter.

ABSTRACT

The view of a node in a port-labeled network is an infinite tree encoding all walks in the network originating from this node. We prove that for any integers $n\geq D\geq 1$, there exists a port-labeled network with at most $n$ nodes and diameter at most $D$ which contains a pair of nodes whose (infinite) views are different, but whose views truncated to depth $Ω(D\log (n/D))$ are identical.

Motivation & Objective

• To resolve the tightness of the O(D log(n/D)) upper bound for view distinction in anonymous, port-labeled networks.
• To demonstrate that Ω(D log(n/D)) depth is necessary to distinguish nodes with different infinite views.
• To construct explicit network families where truncated views up to O(D log(n/D)) depth are identical, yet infinite views differ.
• To close the gap between known upper bounds and lower bounds for view truncation depth in symmetric network topologies.

Proposed method

• Construct a base graph family G_l with diameter 3, containing two nodes a_l and b_l whose views are identical up to depth l = Θ(log n), but differ infinitely.
• Use a recursive construction of G_l based on symmetric port labeling and path isomorphism to ensure view agreement up to a logarithmic depth.
• Subdivide edges in G_l by inserting D'/3 nodes per edge to scale the diameter to D' while preserving the view agreement up to depth Θ(D' log(n'/D')).
• Apply a transformation ξ_D(G_l) that replaces each edge with a path of length D, preserving port labeling and view structure up to depth D(l-1).
• Prove via structural induction and path isomorphism that views of a_l and b_l are identical up to depth D(l-1), but differ infinitely due to structural asymmetry in the base graph.
• Establish the lower bound by selecting parameters such that the depth D(l-1) reaches (D'-5)/6 log₂(n'/D') - 0.41D' for n' ≥ D'·2¹²/3.

Experimental results

Research questions

• RQ1 Is the O(D log(n/D)) upper bound for view distinction in anonymous networks asymptotically tight?
• RQ2 Can a pair of nodes in a network with diameter D and n nodes have identical views up to depth O(D log(n/D)) yet different infinite views?
• RQ3 What is the minimal depth l such that identical truncated views up to depth l guarantee identical infinite views in symmetric networks?
• RQ4 Can such a lower bound be constructed explicitly for arbitrary D and n with D ≤ n?
• RQ5 Does the bound Ω(D log(n/D)) hold even when the network is required to be connected and port-labeled?

Key findings

• For any D' ≥ 3 and n' ≥ D'·2¹²/3, there exists a network with at most n' nodes and diameter at most D' containing two nodes with identical views truncated to depth (D'-5)/6 log₂(n'/D') - 0.41D' but different infinite views.
• The constructed network achieves a lower bound of Ω(D log(n/D)) on the depth required to distinguish nodes with different views, matching the best-known upper bound.
• The construction relies on a base graph G_l with diameter 3 and two nodes a_l, b_l whose views agree up to depth l = Θ(log n), but diverge infinitely.
• The edge subdivision technique ξ_D(G_l) scales the diameter to D' while preserving view agreement up to depth Θ(D' log(n'/D')) for the scaled network.
• The bound is tight: no smaller depth suffices to distinguish all pairs of nodes with different views in symmetric anonymous networks.
• The result resolves the open problem posed by Hendrickx (2014) regarding the tightness of the O(D log(n/D)) bound for view distinction.